# Ollama hosting

Ollama runs AI models in your own environment. Use a VPS to experiment with models that fit its CPU and memory resources.

- Test a local model
- Connect an application to a model API
- Compare memory needs of lightweight models

These VPS plans do not include a GPU. Check RAM, model size and acceptable latency before choosing a plan. Each model has its own usage terms.

## Experiment with a local model

A first project might test drafting or short-text classification with a compatible model. You can then connect your application to the model API. Evaluate results on your own examples before choosing a business use: quality varies with the model and instructions.

## Measure before sizing

Start with a lightweight model and measure memory use and response times. These VPS offers are not GPU servers. Concurrent requests and context length can significantly change requirements; check the selected model’s licence too.

## Can I run any model?

No. Model size and context can exceed server resources. Validate a suitable model before ordering; no generation speed is guaranteed.